Claude Code v2.1.163: Version Pinning, Plugin List, Hook Improvements, and Critical Bug Fixes

Claude Code v2.1.163 ships with several managed settings additions, new commands, and a batch of important bug fixes. Here's what changed.
Managed Settings: Version Pinning
Added requiredMinimumVersion and requiredMaximumVersion managed settings. If the Claude Code version falls outside the allowed range, it refuses to start and directs users to an approved version. Useful for org-wide compliance and ensuring all developers run a controlled version.
New Command: /plugin list
Introduced /plugin list to list installed plugins, with --enabled and --disabled filters. No more digging through config files to see which plugins are active.
Hook Improvement: Additional Context
Stop and SubagentStop hooks can now return hookSpecificOutput.additionalContext to pass feedback to Claude without triggering a hook error. This allows hooks to keep the conversation turn going by supplying extra context.
Skills: Dollar Sign Escaping
Added \$ escape syntax to include a literal $ before a digit in command bodies. Previously, such sequences were interpreted as variable expansion.
MCP Servers: Session ID on Resume
stdio MCP servers now receive the same CLAUDE_CODE_SESSION_ID as hooks/Bash when resuming with --resume.
Bug Fixes
- claude -p hang: Fixed foreground
claude -phanging forever if a backgrounded command never exits. Background shells are now stopped ~5 seconds after the final result once stdin closes. - Bedrock/Vertex/Foundry + CI: Fixed
claude -pfailing withANTHROPIC_API_KEY requiredon these platforms whenCI=trueand no Anthropic API key is set. - Bazel/EDR workflows: Fixed bash command failures where
$TMPDIRwas incorrectly overridden to/tmp/claude-{uid}for all commands instead of only sandboxed ones (regression in 2.1.154). - Windows EEXIST: Fixed bash commands failing on Windows with
EEXIST: file already existson the session-env directory when it has the read-only attribute or is inside OneDrive. - Permission rules timing: Fixed org-managed permission rules not applying for the entire session when the managed settings fetch completed during startup on a fresh config directory.
- Background sessions after update: Fixed background sessions in
claude agentslosing their running background tasks when reattached after a Claude Code update. Also, background agent sessions now update to a new version in the background, so opening a session after an update no longer waits on a cold restart. - Terminal/UI glitches: Fixed terminal misalignment and multi-second hang when exiting agent view by pressing Esc; fixed Stop button on background-task chip not clearing when the process was already gone; fixed keyboard input becoming permanently unresponsive after a paste operation whose end marker was dropped.
- Hook condition matching: Fixed hook
if: "Bash(...)"conditions firing on every Bash command containing$()or$VAR. The pattern now also matches commands inside subshells and backticks. - Deny rules with $HOME: Fixed deny rules on home-directory paths (e.g.,
Read(~/Desktop/**)) not blocking Bash commands that reference the path via$HOME. - Transcript artifact: Fixed a stray "(no content)" line left in the transcript after closing panel dialogs like
/mcpand/plugins.
UI/UX Improvements
- The
/menu now has clearer descriptions for built-in commands and skills. - The subscription-switch suggestion now shows in the startup announcement slot instead of a toast.
claude agentsdispatching from the state-grouped view now starts the session in the directory the agent view was opened from.
📖 Read the full source: GitHub Claude-Code
👀 See Also

Hybrid AI Architecture: Open-Source Components with Proprietary Reasoning Models
A practical hybrid AI architecture is emerging where 89% of organizations use open-source components to reduce costs by over 50%, while proprietary models handle complex reasoning tasks. Open-source frameworks offer transparency and fine-tuning capabilities without licensing negotiations.

Exploring Clawra's Architecture and Social Autonomy Framework
David Im's Clawra experiments with a parallel world framework for AI companions, focusing on autonomy and local-first data privacy.
Parameter Golf: OpenAI's AI-Assisted ML Research Experiment
OpenAI ran Parameter Golf, a competition with 1,000+ participants and 2,000+ submissions, testing AI-assisted machine learning, coding agents, quantization, and novel model design under strict constraints.

AI Subscriptions Need a Reliable Meter: A Call for Service Transparency
A Reddit post argues that AI subscriptions should provide a basic service receipt showing what model was actually served, reasoning effort, context handling, and any load management, drawing parallels to weights-and-measures norms.